作者: Michael Gerndt , Jorge S. Cardoso , Vladimir Podolskiy , Anshul Jindal , Paul Staab
DOI:
关键词:
摘要: A memory leak in an application deployed on the cloud can affect availability and reliability of application. Therefore, to identify ultimately resolve it quickly is highly important. However, production environment running cloud, detection a challenge without knowledge or its internal object allocation details. This paper addresses this online leaks cloud-based infrastructure having any by introducing novel machine learning based algorithm Precog. This solely uses one metric i.e system's utilization which for leak. The developed algorithm's accuracy was tested 60 virtual machines manually labeled data provided our industry partner Huawei Munich Research Center found that proposed achieves score 85\% with less than half second prediction time per machine.